CLIP_DESeq_09282019
Bing Shui
11/28/2019
Library Loading for the analysis
suppressMessages(
suppressWarnings(
c(library(data.table),
library(tidyverse),
library(rtracklayer),
library(DESeq2),
library(Biostrings),
library(Rsubread),
library(pheatmap),
library(gridExtra),
library(ggrepel),
library(CLIPanalyze),
library(ggplot2),
library(RColorBrewer),
library(mixOmics),
library(plotly))
)
)## [1] "data.table" "stats" "graphics"
## [4] "grDevices" "utils" "datasets"
## [7] "methods" "base" "forcats"
## [10] "stringr" "dplyr" "purrr"
## [13] "readr" "tidyr" "tibble"
## [16] "ggplot2" "tidyverse" "data.table"
## [19] "stats" "graphics" "grDevices"
## [22] "utils" "datasets" "methods"
## [25] "base" "rtracklayer" "GenomicRanges"
## [28] "GenomeInfoDb" "IRanges" "S4Vectors"
## [31] "BiocGenerics" "parallel" "stats4"
## [34] "forcats" "stringr" "dplyr"
## [37] "purrr" "readr" "tidyr"
## [40] "tibble" "ggplot2" "tidyverse"
## [43] "data.table" "stats" "graphics"
## [46] "grDevices" "utils" "datasets"
## [49] "methods" "base" "DESeq2"
## [52] "SummarizedExperiment" "DelayedArray" "matrixStats"
## [55] "Biobase" "rtracklayer" "GenomicRanges"
## [58] "GenomeInfoDb" "IRanges" "S4Vectors"
## [61] "BiocGenerics" "parallel" "stats4"
## [64] "forcats" "stringr" "dplyr"
## [67] "purrr" "readr" "tidyr"
## [70] "tibble" "ggplot2" "tidyverse"
## [73] "data.table" "stats" "graphics"
## [76] "grDevices" "utils" "datasets"
## [79] "methods" "base" "Biostrings"
## [82] "XVector" "DESeq2" "SummarizedExperiment"
## [85] "DelayedArray" "matrixStats" "Biobase"
## [88] "rtracklayer" "GenomicRanges" "GenomeInfoDb"
## [91] "IRanges" "S4Vectors" "BiocGenerics"
## [94] "parallel" "stats4" "forcats"
## [97] "stringr" "dplyr" "purrr"
## [100] "readr" "tidyr" "tibble"
## [103] "ggplot2" "tidyverse" "data.table"
## [106] "stats" "graphics" "grDevices"
## [109] "utils" "datasets" "methods"
## [112] "base" "Rsubread" "Biostrings"
## [115] "XVector" "DESeq2" "SummarizedExperiment"
## [118] "DelayedArray" "matrixStats" "Biobase"
## [121] "rtracklayer" "GenomicRanges" "GenomeInfoDb"
## [124] "IRanges" "S4Vectors" "BiocGenerics"
## [127] "parallel" "stats4" "forcats"
## [130] "stringr" "dplyr" "purrr"
## [133] "readr" "tidyr" "tibble"
## [136] "ggplot2" "tidyverse" "data.table"
## [139] "stats" "graphics" "grDevices"
## [142] "utils" "datasets" "methods"
## [145] "base" "pheatmap" "Rsubread"
## [148] "Biostrings" "XVector" "DESeq2"
## [151] "SummarizedExperiment" "DelayedArray" "matrixStats"
## [154] "Biobase" "rtracklayer" "GenomicRanges"
## [157] "GenomeInfoDb" "IRanges" "S4Vectors"
## [160] "BiocGenerics" "parallel" "stats4"
## [163] "forcats" "stringr" "dplyr"
## [166] "purrr" "readr" "tidyr"
## [169] "tibble" "ggplot2" "tidyverse"
## [172] "data.table" "stats" "graphics"
## [175] "grDevices" "utils" "datasets"
## [178] "methods" "base" "gridExtra"
## [181] "pheatmap" "Rsubread" "Biostrings"
## [184] "XVector" "DESeq2" "SummarizedExperiment"
## [187] "DelayedArray" "matrixStats" "Biobase"
## [190] "rtracklayer" "GenomicRanges" "GenomeInfoDb"
## [193] "IRanges" "S4Vectors" "BiocGenerics"
## [196] "parallel" "stats4" "forcats"
## [199] "stringr" "dplyr" "purrr"
## [202] "readr" "tidyr" "tibble"
## [205] "ggplot2" "tidyverse" "data.table"
## [208] "stats" "graphics" "grDevices"
## [211] "utils" "datasets" "methods"
## [214] "base" "ggrepel" "gridExtra"
## [217] "pheatmap" "Rsubread" "Biostrings"
## [220] "XVector" "DESeq2" "SummarizedExperiment"
## [223] "DelayedArray" "matrixStats" "Biobase"
## [226] "rtracklayer" "GenomicRanges" "GenomeInfoDb"
## [229] "IRanges" "S4Vectors" "BiocGenerics"
## [232] "parallel" "stats4" "forcats"
## [235] "stringr" "dplyr" "purrr"
## [238] "readr" "tidyr" "tibble"
## [241] "ggplot2" "tidyverse" "data.table"
## [244] "stats" "graphics" "grDevices"
## [247] "utils" "datasets" "methods"
## [250] "base" "CLIPanalyze" "GenomicAlignments"
## [253] "Rsamtools" "GenomicFeatures" "AnnotationDbi"
## [256] "plyr" "ggrepel" "gridExtra"
## [259] "pheatmap" "Rsubread" "Biostrings"
## [262] "XVector" "DESeq2" "SummarizedExperiment"
## [265] "DelayedArray" "matrixStats" "Biobase"
## [268] "rtracklayer" "GenomicRanges" "GenomeInfoDb"
## [271] "IRanges" "S4Vectors" "BiocGenerics"
## [274] "parallel" "stats4" "forcats"
## [277] "stringr" "dplyr" "purrr"
## [280] "readr" "tidyr" "tibble"
## [283] "ggplot2" "tidyverse" "data.table"
## [286] "stats" "graphics" "grDevices"
## [289] "utils" "datasets" "methods"
## [292] "base" "CLIPanalyze" "GenomicAlignments"
## [295] "Rsamtools" "GenomicFeatures" "AnnotationDbi"
## [298] "plyr" "ggrepel" "gridExtra"
## [301] "pheatmap" "Rsubread" "Biostrings"
## [304] "XVector" "DESeq2" "SummarizedExperiment"
## [307] "DelayedArray" "matrixStats" "Biobase"
## [310] "rtracklayer" "GenomicRanges" "GenomeInfoDb"
## [313] "IRanges" "S4Vectors" "BiocGenerics"
## [316] "parallel" "stats4" "forcats"
## [319] "stringr" "dplyr" "purrr"
## [322] "readr" "tidyr" "tibble"
## [325] "ggplot2" "tidyverse" "data.table"
## [328] "stats" "graphics" "grDevices"
## [331] "utils" "datasets" "methods"
## [334] "base" "RColorBrewer" "CLIPanalyze"
## [337] "GenomicAlignments" "Rsamtools" "GenomicFeatures"
## [340] "AnnotationDbi" "plyr" "ggrepel"
## [343] "gridExtra" "pheatmap" "Rsubread"
## [346] "Biostrings" "XVector" "DESeq2"
## [349] "SummarizedExperiment" "DelayedArray" "matrixStats"
## [352] "Biobase" "rtracklayer" "GenomicRanges"
## [355] "GenomeInfoDb" "IRanges" "S4Vectors"
## [358] "BiocGenerics" "parallel" "stats4"
## [361] "forcats" "stringr" "dplyr"
## [364] "purrr" "readr" "tidyr"
## [367] "tibble" "ggplot2" "tidyverse"
## [370] "data.table" "stats" "graphics"
## [373] "grDevices" "utils" "datasets"
## [376] "methods" "base" "mixOmics"
## [379] "lattice" "MASS" "RColorBrewer"
## [382] "CLIPanalyze" "GenomicAlignments" "Rsamtools"
## [385] "GenomicFeatures" "AnnotationDbi" "plyr"
## [388] "ggrepel" "gridExtra" "pheatmap"
## [391] "Rsubread" "Biostrings" "XVector"
## [394] "DESeq2" "SummarizedExperiment" "DelayedArray"
## [397] "matrixStats" "Biobase" "rtracklayer"
## [400] "GenomicRanges" "GenomeInfoDb" "IRanges"
## [403] "S4Vectors" "BiocGenerics" "parallel"
## [406] "stats4" "forcats" "stringr"
## [409] "dplyr" "purrr" "readr"
## [412] "tidyr" "tibble" "ggplot2"
## [415] "tidyverse" "data.table" "stats"
## [418] "graphics" "grDevices" "utils"
## [421] "datasets" "methods" "base"
## [424] "plotly" "mixOmics" "lattice"
## [427] "MASS" "RColorBrewer" "CLIPanalyze"
## [430] "GenomicAlignments" "Rsamtools" "GenomicFeatures"
## [433] "AnnotationDbi" "plyr" "ggrepel"
## [436] "gridExtra" "pheatmap" "Rsubread"
## [439] "Biostrings" "XVector" "DESeq2"
## [442] "SummarizedExperiment" "DelayedArray" "matrixStats"
## [445] "Biobase" "rtracklayer" "GenomicRanges"
## [448] "GenomeInfoDb" "IRanges" "S4Vectors"
## [451] "BiocGenerics" "parallel" "stats4"
## [454] "forcats" "stringr" "dplyr"
## [457] "purrr" "readr" "tidyr"
## [460] "tibble" "ggplot2" "tidyverse"
## [463] "data.table" "stats" "graphics"
## [466] "grDevices" "utils" "datasets"
## [469] "methods" "base"Filtering of all peaks
Load the rds file from all samples CLIP analysis and use MA plot to visuaize that sequence depth are balanced between CLIP and IC libraries, and the peaks that were called.
peak.data<- readRDS("~/OneDrive - Harvard University/Haigis Lab/Projects/Halo-Ago2/Halo-Ago-KRas/Raw Data/HEAP-CLIP/HEAP_09282019/Analysis/CLIP_Analysis/joint_analysis/peakdata.2019-12-04.rds")
plotMA(peak.data$gene.counts.nopeaks,
main = "MA plot for all HEAP vs IC\n in genes outside peaks")
plotMA(peak.data$res.deseq,
main = "MA plot for all HEAP vs IC in peaks,\none-sided test")
Prepare the peak, assign scores to KRas peaks by adjusted p-value.
peaks.all <- peak.data$peaks
peaks.all <- subset(peaks.all, width > 20)
peaks.all <- subset(peaks.all, log2FC > 0)
peaks.all <- keepStandardChromosomes(peaks.all)
score(peaks.all) <- -log10(peaks.all$padj)
length(peaks.all)## [1] 15486summary(width(peaks.all))## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 21.00 52.00 57.00 59.26 63.00 195.00count.threshold <- 10
norm.counts <- counts(peak.data$peak.counts, normalized = TRUE)
norm.counts <- norm.counts[names(peaks.all), ]
colnames(norm.counts)[1:12] <- c(paste0("HVA", 1:6), paste0("HVAK", 1:6))
colnames(norm.counts)[13:24] <- c(paste0("HVA-IC", 1:6), paste0("HVAK-IC", 1:6))
selected.peaks <- (rowMeans(norm.counts[, paste0("HVA", 1:6)]) > count.threshold) |
(rowMeans(norm.counts[, paste0("HVAK", 1:6)]) > count.threshold)
peaks.all <- peaks.all[selected.peaks, ]
length(peaks.all)## [1] 6519summary(width(peaks.all))## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 21.00 52.00 56.00 58.05 61.00 142.00peaks.all.overlaps <- peak.data[[2]]
peaks.all.overlaps <-
peaks.all.overlaps[name %in% names(peaks.all)]
peaks.all.lncRNA <-
peaks.all.overlaps[gene_type %in% c("lincRNA", "antisense",
"processed_transcript")]
peaks.all.lncRNA.arbitrary <- peaks.all.lncRNA[, .(name, gene_name)]
peaks.all.lncRNA.arbitrary <-
unique(peaks.all.lncRNA.arbitrary, by = "name")
peaks.all$lncRNA <- as.character(NA)
peaks.all[peaks.all.lncRNA.arbitrary$name, ]$lncRNA <-
peaks.all.lncRNA.arbitrary$gene_name
peaks.all$annot <- ifelse(is.na(peaks.all$lncRNA),
peaks.all$annot,
"lncRNA")
padj.threshold <- 0.05
filter.peaks <- subset(peaks.all, padj < padj.threshold)
score(filter.peaks) <- -log10(filter.peaks$padj)
length(filter.peaks)## [1] 5751summary(width(filter.peaks))## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 22.00 52.00 56.00 57.78 61.00 140.00# HVA peak filtering
peak.data.hva <-
peak.data$peak.counts
colnames(peak.data.hva) <- c(paste0("HVA", 1:6), paste0("HVAK", 1:6), paste0("HVA-IC", 1:6), paste0("HVAK-IC", 1:6))
peak.data.hva <-
peak.data.hva[names(peaks.all),
c("HVA1", "HVA2", "HVA3", "HVA4", "HVA5", "HVA6",
"HVA-IC1", "HVA-IC2", "HVA-IC3", "HVA-IC4", "HVA-IC5", "HVA-IC6")]
colData(peak.data.hva)$batch <- factor(c('1','1','1','2','2','2','1','1','1','2','2','2'))
design(peak.data.hva) <- ~ batch + condition
peak.data.hva <- DESeq(peak.data.hva)## using pre-existing size factors## estimating dispersions## found already estimated dispersions, replacing these## gene-wise dispersion estimates## mean-dispersion relationship## -- note: fitType='parametric', but the dispersion trend was not well captured by the
## function: y = a/x + b, and a local regression fit was automatically substituted.
## specify fitType='local' or 'mean' to avoid this message next time.## final dispersion estimates## fitting model and testingplotMA(peak.data.hva,
main = "MA plot for HVA over input\nin all peaks",
xlab = "Mean of normalized counts")
res.hva <- results(peak.data.hva)
peaks.all$hva.padj <- as.numeric(NA)
peaks.all$hva.log2FC <- as.numeric(NA)
peaks.all[rownames(res.hva), ]$hva.padj <-
res.hva$padj
peaks.all[rownames(res.hva), ]$hva.log2FC <-
res.hva$log2FoldChange
peaks.hva <- subset(peaks.all,
padj < padj.threshold & log2FC > 0 &
hva.padj < padj.threshold &
hva.log2FC > 0)
score(peaks.hva) <- -log10(peaks.hva$hva.padj)
length(peaks.hva)## [1] 4803length(subsetByOverlaps(peaks.hva, filter.peaks))## [1] 4803summary(width(peaks.hva))## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 22.00 52.00 56.00 57.54 61.00 140.00peak.data.hvak <-
peak.data$peak.counts
colnames(peak.data.hvak) <- c(paste0("HVA", 1:6), paste0("HVAK", 1:6), paste0("HVA-IC", 1:6), paste0("HVAK-IC", 1:6))
peak.data.hvak <-
peak.data.hvak[names(peaks.all),
c("HVAK1", "HVAK2", "HVAK3", "HVAK4", "HVAK5", "HVAK6",
"HVAK-IC1", "HVAK-IC2", "HVAK-IC3", "HVAK-IC4", "HVAK-IC5", "HVAK-IC6")]
colData(peak.data.hvak)$batch <- factor(c('1','1','1','2','2','2','1','1','1','2','2','2'))
design(peak.data.hvak) <- ~ batch + condition
peak.data.hvak <- DESeq(peak.data.hvak)## using pre-existing size factors## estimating dispersions## found already estimated dispersions, replacing these## gene-wise dispersion estimates## mean-dispersion relationship## final dispersion estimates## fitting model and testingplotMA(peak.data.hvak,
main = "MA plot for HVAK over input\nin all peaks",
xlab = "Mean of normalized counts")
res.hvak <- results(peak.data.hvak)
peaks.all$hvak.padj <- as.numeric(NA)
peaks.all$hvak.log2FC <- as.numeric(NA)
peaks.all[rownames(res.hvak), ]$hvak.padj <-
res.hvak$padj
peaks.all[rownames(res.hvak), ]$hvak.log2FC <-
res.hvak$log2FoldChange
peaks.hvak <- subset(peaks.all,
padj < padj.threshold & log2FC > 0 &
hvak.padj < padj.threshold &
hvak.log2FC > 0)
score(peaks.hvak) <- -log10(peaks.hvak$hvak.padj)
length(peaks.hvak)## [1] 5491length(subsetByOverlaps(peaks.hvak, filter.peaks))## [1] 5491summary(width(peaks.hvak))## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 22.00 52.00 56.00 57.67 61.00 140.00# for genes inside peaks
colnames(peak.data$peak.counts)[1:12] <- c(paste0("HVA", 1:6), paste0("HVAK", 1:6))
dds.hvak.hva <-
peak.data$peak.counts[names(filter.peaks),
c(paste0("HVA", 1:6), paste0("HVAK", 1:6))]
dds.hvak.hva$sampletype <-
factor(c(rep("HVA", 6), rep("HVAK", 6)), levels = c("HVA", "HVAK"))
dds.hvak.hva$batch <-
factor(c("1","1","1","2","2","2","1","1","1","2","2","2"))
design(dds.hvak.hva) <- ~ batch + sampletype
dds.hvak.hva <- DESeq(dds.hvak.hva)## using pre-existing size factors## estimating dispersions## found already estimated dispersions, replacing these## gene-wise dispersion estimates## mean-dispersion relationship## -- note: fitType='parametric', but the dispersion trend was not well captured by the
## function: y = a/x + b, and a local regression fit was automatically substituted.
## specify fitType='local' or 'mean' to avoid this message next time.## final dispersion estimates## fitting model and testingres.hvak.hva <- results(dds.hvak.hva, contrast = c("sampletype", "HVAK", "HVA"))
res.hvak.hva <- lfcShrink(dds.hvak.hva, contrast = c("sampletype", "HVAK", "HVA"), res = res.hvak.hva, type = "normal")## using 'normal' for LFC shrinkage, the Normal prior from Love et al (2014).
##
## Note that type='apeglm' and type='ashr' have shown to have less bias than type='normal'.
## See ?lfcShrink for more details on shrinkage type, and the DESeq2 vignette.
## Reference: https://doi.org/10.1093/bioinformatics/bty895plotMA(res.hvak.hva,
main = "MA plot for HVAK over HVA\nin all peaks (significant over input)",
xlab = "Mean of normalized counts")
summary(res.hvak.hva)##
## out of 5751 with nonzero total read count
## adjusted p-value < 0.1
## LFC > 0 (up) : 1816, 32%
## LFC < 0 (down) : 103, 1.8%
## outliers [1] : 21, 0.37%
## low counts [2] : 0, 0%
## (mean count < 6)
## [1] see 'cooksCutoff' argument of ?results
## [2] see 'independentFiltering' argument of ?results# for genes outside of peaks
colnames(peak.data$gene.counts.nopeaks)[1:12] <- c(paste0("HVA", 1:6), paste0("HVAK", 1:6))
dds.hvak.hva.peakout <-
peak.data$gene.counts.nopeaks[ ,c(paste0("HVA", 1:6), paste0("HVAK", 1:6))]
dds.hvak.hva.peakout$sampletype <-
factor(c(rep("HVA", 6), rep("HVAK", 6)), levels = c("HVA", "HVAK"))
dds.hvak.hva.peakout$batch <-
factor(c("1","1","1","2","2","2","1","1","1","2","2","2"))
design(dds.hvak.hva.peakout) <- ~ batch + sampletype
dds.hvak.hva.peakout <- DESeq(dds.hvak.hva.peakout)## using pre-existing size factors## estimating dispersions## found already estimated dispersions, replacing these## gene-wise dispersion estimates## mean-dispersion relationship## final dispersion estimates## fitting model and testingres.hvak.hva.peakout <- results(dds.hvak.hva.peakout, contrast = c("sampletype", "HVAK", "HVA"))
res.hvak.hva.peakout <- lfcShrink(dds.hvak.hva.peakout, contrast = c("sampletype", "HVAK", "HVA"), res = res.hvak.hva.peakout, type = "normal")## using 'normal' for LFC shrinkage, the Normal prior from Love et al (2014).
##
## Note that type='apeglm' and type='ashr' have shown to have less bias than type='normal'.
## See ?lfcShrink for more details on shrinkage type, and the DESeq2 vignette.
## Reference: https://doi.org/10.1093/bioinformatics/bty895plotMA(res.hvak.hva.peakout,
main = "MA plot for HVAK over HVA outside of peaks",
xlab = "Mean of normalized counts")
dds_transform <- varianceStabilizingTransformation(dds.hvak.hva)## -- note: fitType='parametric', but the dispersion trend was not well captured by the
## function: y = a/x + b, and a local regression fit was automatically substituted.
## specify fitType='local' or 'mean' to avoid this message next time.rawCountTable_transform <- as.data.frame(assay(dds_transform))
pseudoCount_transform = log2(rawCountTable_transform + 1)
mat.dist = pseudoCount_transform
mat.dist = as.matrix(dist(t(mat.dist)))
mat.dist = mat.dist/max(mat.dist)
setwd("/Users/mizuhi/OneDrive - Harvard University/Haigis Lab/Projects/Halo-Ago2/Halo-Ago-KRas/Raw Data/HEAP-CLIP/HEAP_09282019/Analysis/CLIP_Analysis/Data Visualization")
png('Hierchical_Clustering.png')
cim(mat.dist, symkey = FALSE, margins = c(6, 6))
suppressMessages(dev.off())## quartz_off_screen
## 2Final output is following: 
plotPCA(dds_transform, intgroup = "sampletype", ntop = 500) +
geom_text(aes(label=name), vjust = 2, hjust = -0.1) + xlim(-50,60) + ylim(-40,50)
I would like to just do PCA on samples from HEAP-CLIP_09282019.
dds_09282019 <- dds_transform[, c(paste0("HVA",4:6), paste0("HVAK",4:6))]
plotPCA(dds_09282019, intgroup = "sampletype", ntop = 500) +
geom_text(aes(label=name), vjust = 2, hjust = -0.1)
peaks.all$hvak.hva.padj <- as.numeric(NA)
peaks.all$hvak.hva.log2FC <- as.numeric(NA)
peaks.all[rownames(res.hvak.hva), ]$hvak.hva.padj <-
res.hvak.hva$padj
peaks.all[rownames(res.hvak.hva), ]$hvak.hva.log2FC <-
res.hvak.hva$log2FoldChangeSave the datasets
saveRDS(peaks.all, "Datafiles/peaks-all-09282019.rds")
filter.peaks <- subset(peaks.all, padj < padj.threshold)
saveRDS(filter.peaks, "Datafiles/peaks-filtered-09282019.rds")Map seed to peaks
mirna.info.family <- readRDS("mirna-info-family-seedmatches.rds")
assignMirToPeaks <- function(miRNA = mirs,
peaks = es.peaks.utr3,
database = mirna.info.family){
require(BSgenome)
require(CLIPanalyze)
require(Biostrings)
bsgenome <- load.bsgenome("mm10")
peaks.seq <- get.seqs(bsgenome, peaks)
peaks$seed.8mer <- as.character(NA)
peaks$seed.7m8 <- as.character(NA)
peaks$seed.7A1 <- as.character(NA)
peaks$seed.6mer <- as.character(NA)
for (i in 1:length(miRNA)){
mir <- miRNA[i]
#Prepare seed matches
mir.6m <- as.character(database[database$miR.family %in% mir, ]$seedmatch.6)
mir.7m8 <- as.character(database[database$miR.family %in% mir, ]$seedmatch.m8)
mir.7mA <- as.character(database[database$miR.family %in% mir, ]$seedmatch.A1)
mir.8m <- as.character(database[database$miR.family %in% mir, ]$seedmatch.8)
#Filter peaks with 6mer matches
match <- GRanges(vmatchPattern(mir.6m, peaks.seq))
#Go back to peaks and map the seed match and extend 1 nt on both directions
match.extend <- peaks[seqnames(match),]
match.strand <- as.logical(strand(match.extend) == "+")
start(match.extend[match.strand,]) <- start(match.extend[match.strand,]) + start(match[match.strand,]) - 2
match.extend[match.strand,] <- resize(match.extend[match.strand,], fix = "start", width = 8)
end(match.extend[!match.strand,]) <- end(match.extend[!match.strand,]) - start(match[!match.strand,]) + 2
match.extend[!match.strand,] <- resize(match.extend[!match.strand,], fix = "start", width = 8)
#Assign seed types to these matches
match.extend.seq <- get.seqs(bsgenome, match.extend)
match.extend.seq.df <- data.frame(Peaks = names(match.extend.seq),
Sequence = as.character(match.extend.seq))
for (j in 1:nrow(match.extend.seq.df)){
peak.name <- as.character(match.extend.seq.df[j, "Peaks"])
seq <- as.character(match.extend.seq.df[j, "Sequence"])
if (grepl(mir.8m, seq)){
peaks[peak.name, ]$seed.8mer <- ifelse(is.na(peaks[peak.name, ]$seed.8mer),
mir,
paste(peaks[peak.name, ]$seed.8mer, mir, sep = ", "))
} else if (grepl(mir.7m8, seq)){
peaks[peak.name, ]$seed.7m8 <- ifelse(is.na(peaks[peak.name, ]$seed.7m8),
mir,
paste(peaks[peak.name, ]$seed.7m8, mir, sep = ", "))
} else if (grepl(mir.7mA, seq)){
peaks[peak.name, ]$seed.7A1 <- ifelse(is.na(peaks[peak.name, ]$seed.7A1),
mir,
paste(peaks[peak.name, ]$seed.7A1, mir, sep = ", "))
} else {
peaks[peak.name, ]$seed.6mer <- ifelse(is.na(peaks[peak.name, ]$seed.6mer),
mir,
paste(peaks[peak.name, ]$seed.6mer, mir, sep = ", "))
}
}
}
return(peaks)
}miRNAs with mean counts larger than 200 are selected and mapped to peaks
mirna.family.DGE <- readRDS("Datafiles/mirna-counts-deseq-by-family-09282019.rds")
mirs <- subset(mirna.family.DGE, baseMean > 200)
mirs <- mirs$miR.family
#peaks.mirs <- assignMirToPeaks(miRNA = mirs,
# peaks = filter.peaks,
# database = mirna.info.family)
#saveRDS(peaks.mirs, "Datafiles/peaks-mirs-200-09282019.rds")
peaks.mirs <- readRDS("Datafiles/peaks-mirs-200-09282019.rds")targetofmiR <- function(peaks.mir = brain.peaks.mirs,
miRNA = "",
sitetype = "8mer"){
peaks.mir.sub <- as.data.frame(peaks.mir[,c("log2FC", "padj",
"seed.8mer", "seed.7m8", "seed.7A1", "seed.6mer")])
peaks.seedmatch <- lapply(c("seed.8mer", "seed.7m8", "seed.7A1", "seed.6mer"),
function(seed){
map <- peaks.mir.sub[grepl(miRNA, peaks.mir.sub[,seed]),]
map <- rownames(map)
map
})
names(peaks.seedmatch) <- c("seed.8mer", "seed.7m8", "seed.7A1", "seed.6mer")
if (sitetype == "8mer"){
maps <- peaks.seedmatch[[1]]
} else if (sitetype == "7mer_above"){
maps <- unique(unlist(peaks.seedmatch[1:3]))
} else if (sitetype == "7mer"){
maps <- unique(unlist(peaks.seedmatch[2:3]))
} else if (sitetype == "6mer"){
maps <- peaks.seedmatch[[4]]
} else {
print("Please input site type as: 8mer, 7mer_above, 7mer or 6mer")
}
return(peaks.mir[maps])
}mirs.peaks <- lapply(mirs,
function(mir){
targetofmiR(miRNA = mir,
peaks.mir = peaks.mirs,
sitetype = "7mer_above")
})
names(mirs.peaks) <- mirs
saveRDS(mirs.peaks, "Datafiles/miRNA-peaks-list-09282019.rds")lens <- as.data.frame(sapply(mirs.peaks, function(x) length(x)))
mirs.peaks.log2FC.median <- sapply(mirs.peaks,
function(list){
log2FCs <- list$hvak.hva.log2FC
return(median(log2FCs))
})
mirs.peaks.log2FC.mean <- sapply(mirs.peaks,
function(list){
log2FCs <- list$hvak.hva.log2FC
return(mean(log2FCs))
})
mirs.targets.log2FC <- cbind(as.data.frame(mirs.peaks.log2FC.median),
as.data.frame(mirs.peaks.log2FC.mean),
lens)
colnames(mirs.targets.log2FC) <- c("targets_log2FC_median","targets_log2FC_mean", "N")
mirs.targets.log2FC <- merge(mirs.targets.log2FC, mirna.family.DGE[,c("log2FoldChange", "padj")], by = 0)
colnames(mirs.targets.log2FC)[1] <- "miR.family"Show any miRNA that has more than 20 peaks matched
df <- subset(mirs.targets.log2FC, N > 20)
p <- ggplot(df, aes(x = log2FoldChange, y = targets_log2FC_median, label = miR.family, size = N)) +
geom_point(colour = "#1C75BB", alpha = 0.6) +
#scale_color_manual(fill = "yellow") +
geom_hline(yintercept = 0, linetype = "dashed", size = 0.3) +
geom_vline(xintercept = 0, linetype = "dashed", size = 0.3) +
xlab("miRNA expression Fold change log2 (HVAK / HVA)") +
ylab("Median of peak signal changes log2 (HVAK / HVA)") +
theme_bw() +
theme(panel.border = element_blank(),
panel.background = element_blank(),
panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
axis.title.x = element_text(size=12, margin = margin(t = 10)),
axis.title.y = element_text(size=12, margin = margin(r = 10)),
axis.text = element_text(size=10),
axis.line.y = element_line(size = 0.5),
axis.line.x = element_line(size = 0.5),
axis.ticks.x = element_line(size = 0),
axis.ticks.y = element_line(size = 0.5),
plot.title = element_text(hjust = 0.5, size = 12, face = "bold"))
p <- ggplotly(p)
pp2 <- ggplot(df, aes(x = log2FoldChange, y = targets_log2FC_mean, label = miR.family, size = N)) +
geom_point(colour = "#EC469A", alpha = 0.6) +
#scale_color_manual(fill = "yellow") +
geom_hline(yintercept = 0, linetype = "dashed", size = 0.3) +
geom_vline(xintercept = 0, linetype = "dashed", size = 0.3) +
xlab("miRNA expression Fold change log2 (HVAK / HVA)") +
ylab("Mean of peak signal changes log2 (HVAK / HVA)") +
theme_bw() +
theme(panel.border = element_blank(),
panel.background = element_blank(),
panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
axis.title.x = element_text(size=12, margin = margin(t = 10)),
axis.title.y = element_text(size=12, margin = margin(r = 10)),
axis.text = element_text(size=10),
axis.line.y = element_line(size = 0.5),
axis.line.x = element_line(size = 0.5),
axis.ticks.x = element_line(size = 0),
axis.ticks.y = element_line(size = 0.5),
plot.title = element_text(hjust = 0.5, size = 12, face = "bold"))
p2 <- ggplotly(p2)
p2If we only look at the top 40 highly expressed miRNAs
mirna.family.DGE <- mirna.family.DGE[order(-mirna.family.DGE$baseMean),]
mirs.40 <- rownames(mirna.family.DGE)[1:40]
df.40 <- df[df$miR.family %in% mirs.40,]p3 <- ggplot(df.40, aes(x = log2FoldChange, y = targets_log2FC_median, label = miR.family, size = N)) +
geom_point(colour = "#1C75BB", alpha = 0.6) +
#scale_color_manual(fill = "yellow") +
geom_hline(yintercept = 0, linetype = "dashed", size = 0.3) +
geom_vline(xintercept = 0, linetype = "dashed", size = 0.3) +
xlab("Top 40 miRNA expression Fold change log2 (HVAK / HVA)") +
ylab("Median of peak signal changes log2 (HVAK / HVA)") +
theme_bw() +
theme(panel.border = element_blank(),
panel.background = element_blank(),
panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
axis.title.x = element_text(size=12, margin = margin(t = 10)),
axis.title.y = element_text(size=12, margin = margin(r = 10)),
axis.text = element_text(size=10),
axis.line.y = element_line(size = 0.5),
axis.line.x = element_line(size = 0.5),
axis.ticks.x = element_line(size = 0),
axis.ticks.y = element_line(size = 0.5),
plot.title = element_text(hjust = 0.5, size = 12, face = "bold"))
p3 <- ggplotly(p3)
p3p4 <- ggplot(df.40, aes(x = log2FoldChange, y = targets_log2FC_mean, label = miR.family, size = N)) +
geom_point(colour = "#EC469A", alpha = 0.6) +
#scale_color_manual(fill = "yellow") +
geom_hline(yintercept = 0, linetype = "dashed", size = 0.3) +
geom_vline(xintercept = 0, linetype = "dashed", size = 0.3) +
xlab("Top 40 miRNA expression Fold change log2 (HVAK / HVA)") +
ylab("Mean of peak signal changes log2 (HVAK / HVA)") +
theme_bw() +
theme(panel.border = element_blank(),
panel.background = element_blank(),
panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
axis.title.x = element_text(size=12, margin = margin(t = 10)),
axis.title.y = element_text(size=12, margin = margin(r = 10)),
axis.text = element_text(size=10),
axis.line.y = element_line(size = 0.5),
axis.line.x = element_line(size = 0.5),
axis.ticks.x = element_line(size = 0),
axis.ticks.y = element_line(size = 0.5),
plot.title = element_text(hjust = 0.5, size = 12, face = "bold"))
p4 <- ggplotly(p4)
p4Here are the top 10 miRNAs with the most peaks associated.
color.vec <- brewer.pal(name = "Spectral", n = 11)
df <- as.data.frame(filter.peaks)
df$hvak.hva.logP <- -log10(df$hvak.hva.padj)
len <- sapply(mirs.peaks, function(x) length(x))
mirs.peaks <- mirs.peaks[order(-len)]
mirna <- names(mirs.peaks)
mirs.peaks.names <- lapply(mirs.peaks, function(x) names(x))
for (i in 1:10){
p <- ggplot() +
geom_point(data = df, aes(x = hvak.hva.log2FC, y = hvak.hva.logP), size = 1, alpha = 0.5, color = "grey80") +
geom_point(data = df[mirs.peaks.names[[mirna[i]]],], aes(x = hvak.hva.log2FC, y = hvak.hva.logP), size = 1, alpha = 0.7, color = "#7570B3") +
geom_hline(yintercept = -log10(0.05), linetype = "dashed", size = 0.3) +
geom_vline(xintercept = c(-0.5, 0.5), linetype = "dashed", size = 0.3) +
xlab("Fold change log2 (HVAK / HVA)") +
ylab("-log10(FDR)") +
theme_bw() +
theme(panel.border = element_blank(),
panel.background = element_blank(),
panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
axis.title.x = element_text(size=12, margin = margin(t = 10)),
axis.title.y = element_text(size=12, margin = margin(r = 10)),
axis.text = element_text(size=10),
axis.line.y = element_line(size = 0.5),
axis.line.x = element_line(size = 0.5),
axis.ticks.x = element_line(size = 0),
axis.ticks.y = element_line(size = 0.5),
plot.title = element_text(hjust = 0.5, size = 12, face = "bold")) +
#guides(fill = guide_legend(title = "miRNA family")) +
ggtitle(sprintf("%s targets", mirna[i ]))
print(p)
}## Warning: Removed 21 rows containing missing values (geom_point).
## Warning: Removed 21 rows containing missing values (geom_point).## Warning: Removed 1 rows containing missing values (geom_point).
## Warning: Removed 21 rows containing missing values (geom_point).
## Warning: Removed 1 rows containing missing values (geom_point).
## Warning: Removed 21 rows containing missing values (geom_point).
## Warning: Removed 1 rows containing missing values (geom_point).
## Warning: Removed 21 rows containing missing values (geom_point).## Warning: Removed 2 rows containing missing values (geom_point).
## Warning: Removed 21 rows containing missing values (geom_point).
## Warning: Removed 2 rows containing missing values (geom_point).
## Warning: Removed 21 rows containing missing values (geom_point).
## Warning: Removed 21 rows containing missing values (geom_point).
## Warning: Removed 21 rows containing missing values (geom_point).## Warning: Removed 1 rows containing missing values (geom_point).
## Warning: Removed 21 rows containing missing values (geom_point).
I just want a table of all top 4o miRNAs and the number of peaks that are associated with them with LFC (HVAK/HVA) > 0 and < 0.
peak.number <- as.data.frame(table(mirs.peaks[[1]]$hvak.hva.log2FC > 0))
peak.number <- t(peak.number[,-1])
colnames(peak.number) <- c("LFC(HVAK/HVA) < 0", "LFC(HVAK/HVA) > 0")
for (i in 2: 40) {
new.peak <- as.data.frame(table(mirs.peaks[[i]]$hvak.hva.log2FC > 0))
new.peak <- t(new.peak[,-1])
peak.number <- rbind(peak.number, new.peak)
}
rownames(peak.number) <- names(mirs.peaks)[1:40]
peak.number <- rbind(peak.number, c(sum(df$hvak.hva.log2FC < 0), sum(df$hvak.hva.log2FC > 0)))
rownames(peak.number)[dim(peak.number)[1]] <- "Total"
peak.number## LFC(HVAK/HVA) < 0 LFC(HVAK/HVA) > 0
## let-7-5p/miR-98-5p 93 512
## miR-29-3p 28 326
## miR-15-5p/16-5p/195-5p/322-5p/497-5p 26 319
## miR-17-5p/20-5p/93-5p/106-5p 56 287
## miR-200-3p/429-3p 69 261
## miR-24-3p 29 268
## miR-194-5p 25 242
## miR-27-3p 13 212
## miR-103-3p/107-3p 10 189
## miR-196-5p 16 179
## miR-23-3p 27 167
## miR-19-3p 23 161
## miR-26-5p 22 154
## miR-25-3p/32-5p/92-3p/363-3p/367-3p 25 139
## miR-21-3p 18 134
## miR-141-3p 29 119
## miR-484 22 117
## miR-30-5p/384-5p 31 106
## miR-130-3p/301-3p 22 113
## miR-22-3p 15 106
## miR-96-5p 25 96
## miR-148-3p/152-3p 34 80
## miR-182-5p 26 84
## miR-181-5p 16 94
## miR-34-5p/449-5p 18 80
## miR-205-5p 16 79
## miR-7-5p 6 87
## miR-31-5p 48 41
## miR-30a-3p/30d-3p/30e-3p 13 70
## miR-132-3p/212-3p 10 70
## miR-674-5p 11 63
## miR-194-2-3p/6926-5p/7055-5p 9 64
## miR-21ac-5p 8 57
## miR-141-5p/6769b-3p 18 41
## miR-423-5p 8 50
## miR-185-5p 12 45
## miR-139-5p 5 52
## miR-671-5p 14 42
## miR-192-5p/215-5p 12 43
## miR-135ab-5p 7 47
## Total 1005 4746